To: Saturn V who wrote (89632 ) 10/7/1999 12:47:00 AM From: Process Boy Read Replies (3) | Respond to of 186894
Saturn - <Cascades with its super large L2 Cache, will also benefit a lot on memory intensive and integer benchmarks. I expect Cascades to outdo K-7 for such applications and for the critical server marketplace. However for floating point, the picture may be muddled. For non-memory intensive floating point benchmarks, the Extra Floating Point Unit may give K-7 an edge , for equivalent MHz. So for superiority on all benchmarks,Intel will have to resort to process optimization or the next generation process to push the Mhz further than AMD. > One more question. What would be your take on a Coppermine with 512 L2? <Intel has sucessfuly concealed Coppermine's potential performance, and may have surprised AMD. But it may also have hurt itself. The K-7 infrastructure has gained more momentum than it should have.> I don't disagree with you here. Intel wil have to continue to perform without major screw ups form this point to use the latest gains to advantage. A successful .18 ramp in four fabs is on the plate now. So far, this item looks good. <Since Intel is playing its cards close to its chest, AMD zealots may be surprised if the 0.13 micron process and Willamette show up sooner than Charles R expects. As a matter of fact Charles R, should break off his association with his friends and sources within Intel. They have caused him great pain where it hurts most. In his pocketbook !> I have keen memories of how Coppermine was "just a Dixon", and much much bally-hooing over the .18 Dixon MHz, and how these were indicators of what was to come. You are right, Intel chooses to play its cards close to its chest. As to the party in question, he claims to have inside Intel sources. If he does, they sure don't know what's going on where I am (Coppermine Product Design and Process Development). PB